Feature Selection in Software Product Line Using Metaheuristic Techniques

Abstract

Software Product Line (SPL) is a part of the software-intensive system which customizes software by combining various existing features of the software with multiple variants. The features are the characteristics of a software system. Software Product Line plays a big role in minimizing cost, utilization of resources and maximizes chances to attain the vision of an organization. Over the past few years, Software Product Line has gained industry attention and acceptance as it helps organizations to achieve and deliver ever evolving customer requirements and needs. The Software Product Line is the mandatory requirement of the software industry as SPL maximizes the possibility of accomplishing the goals. It has multiple applications in the different domains of an organization, to name a few, E-commerce, Airline industry, Mobile software, Spacecraft and Automobile industry. Software Product Line also helps the organization to be competitive by delivering software on time and within cost. The selection of relevant, non-redundant and important features plays a critical role in the improvement of an organization s overall performance. The feature model is used to represent SPL. The feature model represents the feature of a software product along with its relationship. The main challenge is selecting valid features in SPL as different types of dependencies or constraints needs to be addressed. Another challenge with Software Product Line is that combinations of the small number of features can generate an exponential number of product configurations which makes it NP-hard problem. Traditional algorithms are not very successful for applications which have large search space. Therefore, to solve such problems efficiently and for promising results, Hybrid metaheuristic models with improved fitness function have been designed in this study. The results have been divided into five cases. These cases have been used to describe the significance of features with respect to performance. A weight-based methodology is use